Skip to main content

Full text of "USPTO Patents Application 09864621"

See other formats


TOXICITY TYPING USING EMBRYOID BODIES 



CROSS-REFERENCE TO RELATED APPLICATIONS 
This application is a divisional application of application Serial No. 09/457,931, filed 
December 8, 1999, which claims priority under 35 USC § 1 19(e) to U.S. Provisional 
Application Serial No. 60/1 1 1,640, filed December 9, 1998, the entire contents of which are 
incorporated herein by reference. 

TECHNICAL FIELD 
This invention provides methods for identifying and characterizing toxic compounds 
as well as for screening new compounds for toxic effects. 

BACKGROUND ART 

Some 55,000 chemicals are currently produced or used in the United States every 
year. Relatively few of these compounds have undergone comprehensive testing for acute or 
chronic toxicities. One estimate is that less than 1 percent of commercial chemicals have 
undergone a complete health hazard assessment. Faster and less expensive means of testing 
the toxicity of these compounds would be desirable. It would be particularly useful if such 
means were also amenable to high throughput use. 

In addition to industrial and household chemicals, a number of chemical compositions 
are developed each year for use as pharmaceuticals. Rules regarding the testing of potential 
pharmaceuticals are promulgated by the Food and Drug Administration ("FDA"), which 
currently requires comprehensive testing of toxicity, mutagenicity, and other effects in at 
least two species, only one of which can be murine, before a drug candidate can be entered 
into human clinical trials. Preclinical toxicity testing alone costs some hundreds of thousands 
of dollars. 

In 1997, the pharmaceutical industry was estimated to have spent over $4.5 billion on 
screening assays and testing to determine toxicity. Despite this huge investment, almost one 
third of all prospective human therapeutics fail in the first phase of human clinical trials 
because of unexpected toxicity. It is clear that currently available toxicological screening 



1 



assays do not detect all toxicities associated with human therapy. Better means of screening 
potential therapeutics for potential toxicity would reduce the cost and uncertainty of 
developing new therapeutics and, by reducing uncertainty, would encourage the private 
sector to commit additional resources to drug development. 

Currently available alternatives to traditional "single-reporter" cell lines and animal 
toxicity testing do not fully meet these needs. For example, Farr, U.S. Patent 5,81 1,231, 
provides methods of identifying and characterizing toxic compounds by choosing selected 
stress promoters to and determining the level of the transcription of genes linked to these 
promoters in cells of various cell lines. This method therefore depends on the degree to 
which both the promoter and the cell lines are representative of the effect of the potentially 
toxic agent on the organism of interest. 

The use of hybridization arrays of oligonucleotides provides another route for 
determining the potential toxicity of chemical compositions. Exposing cells of a culture to a 
chemical composition and then comparing the expression pattern of the exposed cells to that 
of cells exposed to other chemical agents permits one to detect patterns of expression similar 
to that of the test compound, and thus to predict that the toxicities of the chemical 
compositions will be similar. See, e.g., Service, R., Science 282:396-399 (1998). These 
methods suffer from the fact that individual cell lines may not be fully representative of the 
complex biology of an intact organism. Moreover, even repeating the tests in multiple cell 
lines does not reproduce or account for the complex interactions among cells and tissues that 
occurs in an organism. 

What is needed in the art is a method of systematically testing chemical compositions 
for potential toxicity in a milieu in which cells interact with cells of other types. What is 
further needed is a means of doing so which is relevant to the effect of the composition on 
whole organisms, without the cost, time, and ethical ramification of animal and human 
testing. The present invention addresses these and other needs. 

DISCLOSURE OF THE INVENTION 
This invention provides novel methods for assessing the toxicity of chemical 
compositions. In one group of embodiments, the invention is directed to methods of creating 
a molecular profile of a chemical composition, comprising the steps of a) contacting an 



isolated mammalian embryoid body (EB) with the chemical composition; and b) recording 
alterations in gene expression or protein expression in the mammalian embryoid body in 
response to the chemical composition to create a molecular profile of the chemical 
composition. 

The invention further embodies methods of compiling a library of molecular profiles 
of chemical compositions having predetermined toxicities, comprising the steps of a) 
contacting an isolated mammalian embryoid body with a chemical composition having 
predetermined toxicities; b) recording alterations in gene expression or protein expression in 
the mammalian embryoid body in response to the chemical composition to create a molecular 
profile of the chemical composition; and c) compiling a library of molecular profiles by 
repeating steps a) and b) with at least two chemical compositions having predetermined 
toxicities. 

Another embodiment of the present invention provides methods for typing toxicity of 
a test chemical composition by comparing its molecular profile in EB cells with that of an 
identified chemical composition with predetermined toxicity. In one aspect, the test chemical 
composition can be the same as the chemical composition having predetermined toxicities. 
For example, the test chemical is identified through this testing as exhibiting the identical 
molecular profile as the known chemical composition. 

The invention further encompasses systemic methods for typing the toxicity of a test 
chemical composition by making the profile comparison with a library comprising profiles of 
multiple chemical compositions with predetermined toxicities. Preferably, the chemical 
compositions comprised in a library exert similar toxicities in terms of types and target 
tissues or organs. The library can be in the form of a database. A database may comprise 
more than one library for chemical compositions of different toxicity categories. 

In one aspect of the present invention, the toxicity of a test chemical composition can 
be ranked according to a comparison of its molecular profile in EB cells to those of chemical 
compositions with predetermined toxicities. 

Embryoid bodies in the present invention can be of human or non-human mammals, 
including those of murine species, as well as canine, feline, porcine, bovine, caprine, equine, 
and sheep species. 



The alterations in levels of gene or protein expression can be detected by use of a 
label selected from any of the following: fluorescent, colorimetric, radioactive, enzyme, 
enzyme substrate, nucleoside analog, magnetic, glass, or latex bead, colloidal gold, and 
electronic transponder. The alterations can also be detected by mass spectrometry. The 
chemical composition can be known (for example, a potential new drug) or unknown (for 
example, a sample of an unknown chemical found dumped near a roadside and of unknown 
toxicity). 

Further, the chemical compositions can be therapeutic agents (or potential therapeutic 
agents), of agents of known toxicities, such as neurotoxins, hepatic toxins, toxins of 
hematopoietic cells, myotoxins, carcinogens, teratogens, or toxins to one or more 
reproductive organs. The chemical compositions can further be agricultural chemicals, such 
as pesticides, fungicides, nematicides, and fertilizers, cosmetics, including so-called 
"cosmeceuticals," industrial wastes or by-products, or environmental contaminants. They 
can also be animal therapeutics or potential animal therapeutics. 

The invention further includes integrated systems for comparing the molecular profile 
of a chemical composition to a library of molecular profiles of chemical compositions, 
comprising an array reader adapted to read the pattern of labels on an array, operably linked 
to a computer comprising a data file having a plurality of gene expression or protein 
expression profiles of mammalian embryoid bodies contacted with known or unknown 
chemical compositions. 

The invention also includes integrated systems for correlating the molecular profile 
and toxicity of a chemical composition comprising an array reader adapted to read the pattern 
of labels on an array, operably linked to a digital computer comprising a database file having 
a plurality of molecular profiles of chemical compositions with predetermined toxicities and 
a program suitable for molecular profile-toxicity correlation. The integrated systems of the 
invention can be capable of reading more than 500 labels in an hour, and further can be 
opeably linked to an optical detector for reading the pattern of labels on an array. 



BRIEF DESCRIPTION OF THE DRAWINGS 
Figure 1 depicts differences in expression of nuclear proteins between embryoid 
bodies exposed to one of two drugs, and control embryoid bodies. 



Figure 1 A is a half-tone reproduction of a readout from the mass spectrometer. The 
top band is the mass spectrum for control embryoid bodies, which were grown in the absence 
of either of the test chemical compositions. The middle band is the mass spectrum for the 
embryoid bodies grown in the presence of added troglitazone, and the bottom band of Figure 
1 A shows the mass spectrum of nuclear proteins expressed by embryoid bodies exposed to 
erythromycin estolate. 

Figures IB and 1C are bar graphs that represent computational subtractions of 
identical proteins between the respective test embryoid bodies and the control embryoid 
bodies to indicate only those proteins which are significantly different in expression between 
the test and the control embryoid bodies. Each bar represents a single protein and the height 
of the bar represents the amount of protein expressed by the embryoid bodies exposed to the 
test composition compared to the amount expressed by embryoid bodies not exposed to the 
chemical composition. Figure IB: protein expression of test embryoid bodies contacted with 
troglitazone compared to protein expression of controls. Figure 1C: protein expression of 
test embryoid bodies contacted with erythromycin estolate compared to protein expression of 
controls. 

Figure 2 is a bar graph showing expression of small nuclear proteins detected by mass 
spectrometry. X-axis: mass of protein detected. Y-axis: amount of protein detected, in 
relative units. Figure 2A: Protein expression of control embryoid bodies not exposed to the 
chemical composition. Figure 2B: Protein expression of embryoid bodies exposed to 
troglitazone. Figure 2C: Protein expression of embryoid bodies exposed to erythromycin 
estolate. Bold lines indicate proteins expressed in different amounts between embryoid 
bodies exposed to troglitazone and those exposed to erythromycin estolate. 

Figure 3 is a bar graph showing expression of small cytoplasmic proteins detected by 
mass spectrometry. X-axis: mass of protein detected. Y-axis: amount of protein detected, in 
relative units. Figure 3 A: Protein expression of control embryoid bodies not exposed to the 
chemical composition. Figure 3B: Protein expression of embryoid bodies exposed to 
troglitazone. Figure 3C: Protein expression of embryoid bodies exposed to erythromycin 
estolate. Bold lines indicate proteins expressed in different amounts between embryoid 
bodies exposed to troglitazone and those exposed to erythromycin estolate. 



Figure 4 is a bar graph showing expression of large nuclear proteins detected by mass 
spectrometry. X-axis: mass of protein detected. Y-axis: amount of protein detected, in 
relative units. Figure 4A: Protein expression of control embryoid bodies not exposed to the 
chemical composition. Figure 4B: Protein expression of embryoid bodies exposed to 
troglitazone. Figure 4C: Protein expression of embryoid bodies exposed to erythromycin 
estolate. Bold lines indicate proteins expressed in different amounts between embryoid 
bodies exposed to troglitazone and those exposed to erythromycin estolate. 

MODEfS) FOR CARRYING OUT THE INVENTION 

A. DEFINITIONS 

As used herein, "embryoid body", "EB" or "EB cells" typically refers to a 
morphological structure comprised of a population of cells, the majority of which are derived 
from embryonic stem ("ES") cells that have undergone differentiation. Under culture 
conditions suitable for EB formation (e.g., the removal of Leukemia inhibitory factor or 
other, similar blocking factors), ES cells proliferate and form small mass of cells that begin 
to differentiate. In the first phase of differentiation, usually corresponding to about days 1-4 
of differentiation for humans, the small mass of cells forms a layer of endodermal cells on 
the outer layer, and is considered a "simple embryoid body." In the second phase, usually 
corresponding to about days 3-20 post-differentiation for humans, "complex embryoid 
bodies" are formed, which are characterized by extensive differentiation of ectodermal and 
mesodermal cells and derivative tissues. As used herein, the term "embryoid body" or "EB" 
encompasses both simple and complex embryoid bodies unless otherwise required by 
context. The determination of when embryoid bodies have formed in a culture of ES cells is 
routinely made by persons of skill in the art by, for example, visual inspection of the 
morphology. Floating masses of about 20 cells or more are considered to be embryoid 
bodies. See. e.g., Schmitt, R., et ah (1991) Genes Dev. 5:728-740; Doetschman, T.C., et ah 
(1985) /. Embryol Exp. Morph. 87:27-45. It is also understood that the term "embryoid 
body," "EB," or "EB cells" as used herein encompasses a population of cells, the majority of 
which being pluripotent cells capable of developing into different cellular lineages when 
cultured under appropriate conditions. As used herein, the term also refers to equivalent 



structures derived from primordial germ cells, which are primitive cells extracted from 
embryonic gonadal regions. See, e.g., Shamblott, et al (1998) Proc Natl Acad Sci (USA) 
95:13726-13731. Primordial germ cells, sometimes also referred to in the art as ES cells or 
embryonic germ cells, when treated with appropriate factors form pluripotent ES cells from 
which embryoid bodies can be derived. See, e.g., Hogan, U.S. Patent 5,670,372; Shamblott, 
et al., supra. 

"Toxicity," as used herein, means any adverse effect of a chemical on a living 
organism or portion thereof. The toxicity can be to individual cells, to a tissue, to an organ, 
or to an organ system. A measurement of toxicity is therefore integral to determining the 
potential effects of the chemical on human or animal health, including the significance of 
chemical exposures in the environment. Every chemical, and every drug, has an adverse 
effect at some concentration; accordingly, the question is in part whether a drug or chemical 
poses a sufficiently low risk to be marketed for a stated purpose, or, with respect to an 
environmental contaminant, whether the risk posed by its presence in the environment 
requires special precautions to prevent its release, or quarantining or remediation once it is 
released. See, e.g., Klaassen, et al, eds., Casarett and DoulVs Toxicology: The Basic 
Science of Poisons, McGraw-Hill (New York, NY, 5 th Ed. 1996). As used herein, a 
chemical composition with "predetermined toxicities" means that the type of toxicities and/or 
certain pharmacodynamic properties of the chemical composition have been determined. For 
example, a chemical composition may be known to induce liver toxicity. Furthermore, the 
severity of liver toxicity caused by the chemical may be quantitatively measured by the 
amount or concentration of the chemical in contact with the liver tissues. 

"Alteration in gene or protein expression" according to the present invention means a 
change in the expression level of one or more genes or proteins compared to the gene or 
protein expression level of an embryoid body which has been exposed only to normal tissue 
culture medium and normal culturing conditions. Depending on the context, the phrase can 
mean an alteration in the expression of a single protein or gene, as when an embryoid body 
exposed to a chemical agent expresses a protein not expressed by a control embryoid body, 
or it can mean the overall pattern of protein expression of an embryoid body (or group of 
embryoid bodies). 



"Chemical composition ," "chemical," "composition ," and "agent " as used herein, are 
generally synonymous and refer to a compound of interest. The chemical can be, for 
example, one being considered as a potential therapeutic, an agricultural chemical, an 
environmental contaminant, or an unknown substance found at a crime scene, at a waste 
disposal site, or dumped at the side of a road. 

As used herein, "molecular profile" or "profile" of a chemical composition refers to a 
pattern of alterations in gene or protein expression, or both, in an embryoid body contacted 
by the chemical composition compared to a like embryoid body in contact only with culture 
medium. 

As used herein, "database" refers to an ordered system for recording information 
correlating information about the toxicity, the biological effects, or both, of a chemical agent 
to the alterations in the pattern of gene or protein expression, or both, in an embryoid body 
contacted by a chemical composition compared to a like embryoid body in contact only with 
culture medium. 

A "library," as used herein, refers to a compilation of molecular profiles of at least 
two chemical compositions, permitting a comparison of the alterations in gene or protein 
expression, or both, in an embryoid body contacted by a chemical composition to the profiles 
of such expression(s) caused by other chemical compositions. 

"Array" means an ordered placement or arrangement. Most commonly, it is used 
herein to refer to an ordered placement of oligonucleotides (including cDNAs and genomic 
DNA) or of ligands placed on a chip or other surface used to capture complementary 
oligonucleotides (including cDNAs and genomic DNA) or substrates for the ligand. Since 
the oligonucleotide or ligand at each position in the arrangement is known, the sequence (of a 
nucleic acid) or a physical property (of a protein) can be determined by the position to which 
the nucleic acid or substrate binds to the array. 

"Operably linked" means that two or more elements are connected in a way that 
permits an event occurring in one element (such as a reading by an optical reader) to be 
transmitted to and acted upon by a second element (such as a calculation by a computer 
concerning data from an optical reader). 



8 



B. GENERAL DESCRIPTION 

The invention provides methods of assessing toxicity of chemical compositions on a 
genome-wide basis, in a system that closely models the complex biological and cellular 
interactions of whole organisms, including the human body. In one aspect, the invention is 
especially useful in drug development, both because of its ability to validate targets and 
because of its ability to rapidly identify and to quantify all the expressed genes associated 
with responses to a potential therapeutic agent. 

The invention achieves these goals by exploiting the properties of embryoid bodies. 
Embryoid bodies represent a complex group of cells differentiating into different tissues. In 
one embodiment, the cells within an EMBRYOID BODY are substantially synchronized for 
their differentiation. Accordingly, at known intervals, the majority of the synchronized cells 
differentiate into the three embryonic germ layers and further differentiate into multiple 
tissue types, such as cartilage, bone, smooth and striated muscle, and neural tissue, including 
embryonic ganglia. Thus, the cells within embryoid bodies provide a much closer model to 
the complexity of whole organisms than do traditional single cell or yeast assays, while still 
avoiding the cost and difficulties associated with the use of mice and larger mammals. 
Moreover, the recent availability of human embryoid bodies improves the predictive abilities 
of the invention by providing an even closer vehicle for modeling toxicity in human organ 
systems, and in humans. 

The embryoid body of the invention comprises a cell population, the majority of 
which being pluripotent cells capable of developing into different cellular lineages when 
cultured under appropriate conditions. It is preferred that the embryoid body comprises at 
least 51% pluripotent cells derived from totipotent ES cells. More preferably, the embryoid 
body comprises at least 75% pluripotent cells derived from totipotent ES cells. And still 
more preferably, the embryoid body comprises at least 95% pluripotent cells derived from 
totipotent ES cells. 

In its simplest form, the method of creating a molecular profile according to the 
present invention involves contacting embryoid bodies with a chemical composition of 
interest, and then determining the alterations in gene expression, protein expression, or both, 
in the embryoid body exposed to the chemical composition (the "test embryoid body") 



compared to a embryoid body which was not exposed to the agent (a "control embryoid 
body"). 

Furthermore, a library can be generated by compiling molecular profiles for two or 
more different chemical compositions, such as those having similar toxicities. The molecular 
profiles of these compositions can be compared with each other, either qualitatively or 
quantitatively, in order to discern common alterations in their gene or protein expression 
patterns. For example, while the overall gene or protein expression pattern for each chemical 
composition maybe unique, the changes in expression level of certain specific genes or 
proteins may be similar among compositions having similar toxicities-some genes/proteins 
may be similarly up-regulated and therefore expressed in higher amount compared to 
controls; while other genes/proteins may be similarly down-regulated and therefore 
expressing in smaller amount compared to controls. These common molecular features of 
the chemical compositions can then be correlated to their toxicities and serve as surrogate 
markers for assessing the toxicities of a new or previously untested chemical composition, 
such as a drug lead in drug screening assays. 

Thousands of compounds have undergone preclinical and clinical studies. Preclinical 
studies include, among other things, toxicity studies in at least two mammalian species, one 
of which is usually a murine species, typically mice or rats, and clinical trials always include 
information on any apparent toxicity. A considerable amount of information is available 
about the toxicity of various of these compounds. Based on the toxicity information 
available, these compounds can be classified into particular categories of toxicities. For 
example, a number of chemical compositions are listed in Table 1 according to tissues or 
organs in which they exet toxicities. 



10 



TABLE 1 



TOXICITIES 



Drugs 


Dev 


Liver 


CV CNS 


Blood 


Indication 


Trade Names 


thalidomide 


+ 












methotrexate 


+ 








antineoplastics 




retinoic acid 


+ 








acne 




valproic acid 


+ 








seizures 


Depakene 


acetominophen 




+ 






analgesic 




isoniazid 










antibiotic 




diclofenac (NSAIDS) 










anti-inflammatory 


Voltarern 


bromofenac (NSAIDS) 




+ 






anti-inflammatory 


Duract 


troglitazone 




+ 






diabetes 


Rezulin™ 


rosiglitazone 




ntc 






diabetes 


Avandia™ 


trovaflozacin 




+ 






antibiotic 


Trovan™ 


ciprofloxacin 




ntc 






antibiotic 


Cipro™ 


erythromycin estolate 




_i_ 
i 






cuj.ii.uiu ijv 




pravastatin 




_j_ 






J.1LJ1U lyJ W \->l 11 


Pravachol^M 


atorvastin 




4- 






lirvi/i 1 nwprifi cy 

I1UIU. lUWC'lllig, 


T initcn*TM 


clofibrate 




ntc 






lipid lUWCllIlg 


A frnmi H 










-h 


antip sychotic 


Clozaril 


chl oro amphenicol 








+ 


d.IlllUiUHL' 


Ohlfyrr^mvrptin 
v_xiii\ji yjiixy tin 


doxorubicin 










all LI lit U L/XClO lL\s J 




r\ cm T> rvn i rvi r* i n 






+ 




a n ti n eot>1 as ti c s 




cy closopho sphaimde 






i 




dnimcopiaa UU-b 




Compounds 














carbon tetrachloride 




+ 










cadmium 




+ 










phallodidin 




+ 










ethanol 




+ 










di-methyl formide 




+ 










dichlorethylene 




+ 










lead 














benzo(a)pyrene 






+ 








allylamine 






+ 








methylmercury 














trimethyltin 






+ 








carbon disulfide 






+ 








acrylamide 






+ 








hexachloraphene 














DMSO 




not well studied 









"ntc" = non-toxic 5 limited toxicity, control 

"Dev" = developmental "CV" = cardiovascular 



"CNS" = central nervous system 



In one embodiment of the invention, compositions known for having liver toxicities 
are used for a systematic analysis of their molecular profiles in EB cells. In another 
embodiment, compositions causing toxicities to the cardiovascular system are evaluated for 
their molecular profiles in EB cells. In yet another embodiment of the invention, 



11 



compositions causing toxicities to the neuronal system are evaluated for their molecular 
profiles in EB cells. Alternatively, known or potential drugs for treating a disease of choice 
can be used together in a systematic analysis of their toxicities. In this regard, for example, 
anti-cancer drugs and drug candidates can be screened for their tissue and organ toxicities. 

According to one aspect of the invention, molecular profiles of chemical 
compositions can be correlated to toxicities these agents demonstrated in non-human 
animals, in humans, or in both. By then comparing the expression pattern of an embryoid 
body exposed to a new or previously untested agent to a library of such profiles of expression 
induced by agents of known toxicity, predictions can be made as to the likely type of toxicity 
of the new agent. Furthermore, the toxicity of the new agent, if any, can be ranked among 
the known toxic compositions, providing information for prioritization in drug development. 

In addition to its utility in drug development, the invention also has uses in other 
arenas in which the toxicity of chemical compositions is of concern. Thus, the invention can 
be utilized to assess the toxicity of agricultural chemicals, such as pesticides and fertilizers. 
It can further be used with cosmetics. For example, it can be used to screen candidate 
cosmetics for toxicity prior to moving the compounds into animal studies, thereby potentially 
reducing the number of animals which need to be subjected to procedures such as the Draize 
eye irritancy test. Similarly, the methods of the invention can be applied to agents intended 
for use as "cosmeceuticals," wherein agents which are primarily cosmetic are also asserted 
to have some quasi-therapeutic property. Further, the invention can be used to assess the 
relative toxicity of environmental contaminants, including waste products, petrochemical 
residues, combustion products, and products of industrial processes. Examples of such 
contaminants include dioxins, PCBs, and hydrocarbons. 

In general, it is preferred that the method used to detect the levels of protein or gene 
expression provide at least a relative measure of the amount of protein or gene expression. 
More preferably, the method provides a quantitative measure of protein or gene expression to 
facilitate the comparison of the protein or gene expression of the embryoid bodies exposed to 
the test chemical composition to that of embryoid bodies exposed to chemical compositions 
of known toxicity. 



12 



C. PREPARING EMBRYOID BODIES 

In one embodiment, the embryoid bodies used in the present invention can be derived 
from a population of embryonic stem cells ("ES cells") under culture conditions allowing 
differentiation. ES cells are undifferentiated, immature totipotent cells that are capable of 
giving rise to multiple, specialized cell types and, ultimately, to terminally differentiated 
cells. ES cells are typically derived from the inner cell mass of early blastocysts, and can be 
grown indefinitely in culture. See, e.g., Keller et al, WO 96/16162. ES cells are initially 
totipotent, see, e.g., Hogan, U.S. Patent 5,690,926. Techniques for culturing ES cells are 
well known in the art. See, e.g., Robertson, E., "Embryo-derived Stem Cell Lines" in 
Robertson, E. ed., Teratocarcinomas and ES cells: A practical approach, IRL Press 
(Washington, DC 1987); Hogan, R., et aL, eds., Manipulating the Mouse Embryo: A 
Laboratory Manual, Cold Spring Harbor Laboratory Press, (Cold Spring Harbor, NY, 1986). 

Methods for preparing mammalian embryoid bodies using ES cells are known in the 
art. For example, Keller et al., supra, describes preparing EB cell population by culturing ES 
cells in an embryoid body medium. Typically, ES cells remain at an undifferentiated state in 
the presence of Leukemia inhibitory factor ("LIF"). LIF is described, for example, in 
Gearing, U.S. Patent 5,187,077. In vitro propagation of ES cells using LIF is taught in 
Williams, U.S. Patent 5,166,065. 

To commence differentiation, ES cells are removed from the LIF-containing 
embryonic stem cell medium and re-cultured in medium which does not contain LIF. See, 
Keller, et al, supra, at 13. Generally, the cells are cultured in plasticware which has not 
been treated to promote adherence (such as bacterial-grade plasticware, Teflon™ coated 
plasticware, or other materials known to decrease adherence). The cells then tend to bunch 
up, and the interaction of the ES cells as a mass acts to induce the formation of embryoid 
bodies, which commence differentiating into the three germ layers and further into cells of 
particular tissue types, such as muscle cells, epithelial cells, neuronal cells, and 
hematopoietic cells. Snodgrass, , et al, "Embryonic Stem Cells: Research and Clinical 
Potentials" in Smith and Sacher, eds. Peripheral Blood Stem Cells American Association of 
Blood Banks, Bethesda MD (1993). 

Thomson, WO 96/22362, describes a primate ES cell population that remains 
undifferentiated state indefinitely in the presence of fibroblast feeder cells. Feeder cells are 



13 



cells which have been irradiated to remove their ability to divide, but which provide a 
substrate and various factors supporting the culturing of ES cells. See, e.g., Robertson, 
supra, and Hogan, et al, supra. Primary mouse embryo fibroblast cells are preferred, 
although mouse 3T3 or STO cells can be used. E.g., Hogan, et al, supra; Tadaro and Green 
(1963) J. Cell Biol 17:299; Ware and Axelrad (1972) Virology 50:339. Upon removal from 
the feeder cells, the primate ES cells will differentiate into various cell types and, when 
grown at high densities, form embryoid bodies. See, Thomson, supra; Thomson et al. (1996) 
Biol Reprod. 57:254-259; and Thomson and Marshall (1998) Curr Top Dev Biol 38:133- 
165. Formation of embryoid bodies from ES cells of numerous other mammals, such as pigs, 
have also been reported. See, Shim, et al. (1997) Biol Reprod. 57:1089-95. 

Embryoid bodies obtained according to the present invention can be identified 
visually by their morphology, as known in the art and described in Keller et al, supra. Under 
defined culturing conditions, an embryoid body has a general morphology of tightly packed 
cells or cell aggregate or cell mass, in which individual cells are not easily detectable. The 
number of cells in an embryoid body, which can be estimated by the size of the cell mass and 
the approximate size of individual cells, can range from about 5 to about 2,000, although 
preferably from about 10 to about 100. An even more preferred number of cells in an 
embryoid body is about 20. 

Alternatively, the embryoid bodies obtained according to the present invention can be 
identified by the detection of specific markers such as antibodies specific to a population of 
embryoid body cells at defined stage. For example, Keller et al, supra, describes that a Day- 
4 EB cell population expresses substantially low amounts of Sca-1 , C-kit receptor and Class I 
H-2b and essentially no Thy 1, VLA-4, CD44 and CD45. Thus, the cells in a Day-4 EB have 
substantially the same staining pattern when such cells are stained with antibodies to these 
surface antigens. 

If necessary, embryoid bodies obtained and cultured according to the present 
invention may be isolated from the culture based on their physical or chemical properties 
(such as size, mass, density, specific antigen or gene expression), using methods known in 
the art (such as flow cytometry, cell sorting, filtration or centrifugation). 



14 



In a widely noted recent development, two groups have reported the development of 
ES cells from human blastocysts. See, Thomson et al (1998) Science 282: 1 145-1 147 and 
Shamblott, et al. (1998) Proc Natl Acad Sci (USA) 95:13726-13731. 

In Thomson et al s work, human embryos produced by in vitro fertilization for 
clinical purposes were donated by individuals after informed consent and institutional review 
board approval. The embryos were cultured to the blastocyst stage, inner masses isolated, 
and ES cell lines obtained by essentially the same means previously described (and 
referenced above) for nonhuman primate ES cells. Id. The cells were capable of 
differentiating into derivatives of all three embryonic germ layers., Id. As with other primate 
ES cells, LIF was not sufficient to keep the human ES cells from differentiating in the 
absence of fibroblast feeder cells, but differentiated even in the presence of fibroblast feeder 
cells when grown to confluence and allowed to pile up in the culture dish. Id. 

In Shamblott et al 's work, gonadal ridges and mesenteries containing primordial 
germ cells ("PGCs"), taken from human embryos obtained from terminated pregnancies 5-9 
weeks postfertilization, were cultured on mouse STO fibroblast feeder layers in the presence 
of human recombinant LIF, human recombinant basic fibroblast growth factor, and forskolin. 
Over a period of 7-21 days, the PGCs gave rise to colonies of stem cells which developed 
into embryoid bodies. The embryoid bodies were shown to contain a wide variety of 
differentiated cell types, including derivatives of all three embryonic germ layers. It is 
expected that human embryoid bodies such as those created by Thomson et al and Shamblott 
et ah can be used in the methods of the invention. 

ES cells can also be formed from enucleated cells into which the nucleus of a desired 
human or mammalian cell has been inserted. See, e.g., Robl, et al , International Publication 
Number WO 98/07841. 

The embryoid bodies used to test the chemical composition can be of any vertebrate 
species. The choice of the particular species from which the embryoid body is derived will 
typically reflect a balance of several factors. First, depending on the purpose of the study, 
one or more species may be of particular interest. For example, human embryoid bodies will 
be of particular interest for use with compositions being tested as potential human 
therapeutics, while equine, feline, bovine, porcine, caprine, canine, or sheep embryoid bodies 
may be of more interest for a potential veterinary therapeutic. 



15 



Second, even with respect to testing of human therapeutics, cost and handling 
considerations may dictate that some or all testing be performed with non-human, and even 
non-primate embryoid bodies. Obtaining human ES cells, for example, currently requires not 
only informed consent and institutional review board review, but also very labor intensive 
tending. See, Marshall, Science 282:1014-1015 (November 6, 1998). Obtaining primate 
embryoid bodies, while obviously not entailing the same legal requirements, requires first 
obtaining the primates, and entails significant and costly animal husbandry obligations. 
Accordingly, for much testing, it may be desirable to use embryoid bodies from mice, rats, 
guinea pigs, rabbits, and other readily available, and less expensive, laboratory animals. 

Third, it will often be of value to select a species as to which considerable 
information is available on the toxicity of chemical compositions, so that observed changes 
in gene and protein expression can be correlated to various types of toxicity. For this reason, 
mice and rats are preferred embodiments. Most pre-clinical testing is performed on at least 
one murine species, and there therefore exists a large body of information on the toxicity of 
various compounds on various tissues of mice and on rats. Using embryoid bodies derived 
from mice or rats permits the correlation of the alterations in gene or protein expression in 
the embryoid bodies with the toxicities exhibited by these agents in those species. Embryoid 
bodies of other species commonly used in preclinical testing, such as guinea pigs, rabbits, 
pigs, and dogs, are also preferred for the same reason. Typically, embryoid bodies of these 
species will be used for "first pass" screening, or where detailed information on toxicity in 
humans is not needed, or where a result in a murine or other one of these laboratory species 
has been correlated to a known toxicity or other effect in humans. 

Fourth, although primates are not as widely used in preclinical testing and are often 
more expensive to purchase and to maintain than other laboratory animals, their biochemistry 
and developmental biology is considerably closer to that of humans than those of the more 
common laboratory animals. Embryoid bodies derived from primates is therefore preferred 
for toxicity testing where the study is sufficiently important to justify the additional cost and 
handling considerations. Most preferred are human embryoid bodies, since conclusions 
about the toxicity of agents in these embryoid bodies can be considered the most directly 
relevant to the effect of a chemical composition on humans. It is anticipated that studies in 
primate or human embryoid bodies will be performed to confirm results of toxicity studies in 



16 



embryoid bodies of other species. It is anticipated that human embryoid bodies will be used 
where toxicity in humans is of sufficient interest to warrant undertaking the cost and legal 
hurdles, and will become more preferred over time as the legal barriers to the use of human 
ES cells become less onerous. 

Fifth, with respect to human therapeutics, regulatory agencies generally require 
animal data before human trials can begin; it will generally be desirable to use embryoid 
bodies of species which will be used in the preclinical animal studies. The results of toxicity 
testing in the embryoid bodies can then guide the researcher on the degree and type of 
toxicity to anticipate during the animal trials. Certain animal species are known in the art to 
be better models of human toxicity of different types than are others, and species also differ 
in their ability to metabolize drugs. See, e.g., Williams, Environ Health Perspect. 22:133-138 
(1978); Duncan, Adv Sci 23:537-541 (1967). Thus, the particular species preferred for use in 
a particular preclinical toxicity study may vary according to the intended use of the drug 
candidate. For example, a species which provide a suitable model for a drug intended to 
affect the reproductive system may not be as suitable a model for a drug intended to affect 
the nervous system. Criteria for selecting appropriate species for preclinical testing are well 
known in the art. 

While ES cells from different species can be used in the methods of the invention, in 
general, mammalian cells are preferred. In the discussions below, it is assumed that in any 
given comparison of control and test embryoid bodies, the embryoid bodies used as controls 
and those used to test the effects of the chemical compositions are derived from ES cells of 
the same species. 

D. CONTACTING EMBRYOID BODIES WITH CHEMICAL COMPOSITIONS 
1. General 

Once an embryoid body culture has been initiated, it can be contacted with a chemical 
composition. Conveniently, the chemical composition is in an aqueous solution and is 
introduced to the culture medium. The introduction can be by any convenient means, but 
will usually be by means of a pipette, a micropipettor, or a syringe. In some applications, 
such as high throughput screening, the chemical compositions will be introduced by 
automated means, such as automated pipetting systems, which may be on robotic arms. 



17 



Chemical compositions can also be introduced into the medium as in powder or solid forms, 
with or without pharmaceutical excipients, binders, and other materials commonly used in 
pharmaceutical compositions, or with other carriers which might be employed in the intended 
use. For example, chemical compositions intended for use as agricultural chemicals or as 

5 petrochemical agents can be introduced into the medium by themselves to test the toxicity of 

those chemicals or agents, or introduced in combination with other materials with which they 
might be used or which might be found in the environment, to determine if the combination 
of the chemicals or agents has a synergistic effect. Typically, the cultures will be shaken at 
least briefly after introduction of a chemical composition to ensure the composition is 

1 0 dispersed throughout the medium. 

2. Timing of contacting 

% The time as which a chemical composition is added to the culture is within the 

ff discretion of the practitioner and will vary with the particular study objective. Conveniently, 
jp the chemical composition will be added as soon as the embryoid body develops from the 
1 Si I stem cells, permitting the determination of the alteration in protein or gene expression on the 
development of all the tissues of the embryoid body. It may be of interest, however, to focus 
£3 the study on the effect of the composition on a particular tissue type. As previously noted, 
% I individual tissues, such as muscle, nervous, and hepatic tissue, are known to develop at 
W specific times after the embryoid body has formed. Addition of the chemical composition 
2(£ can therefore be staged to occur at the time the tissue of interest commences developing, or at 
a chosen time after commencement of that development, in order to observe the effect on 
altering gene or protein expression in the tissue of interest. 

3. Dosing of the chemical composition 

Different amounts of a chemical composition will be used to contact an embryoid 
25 body depending on the amount of information known about the cytotoxicity of that 

composition, the purposes of the study, the time available, and the resources of the 
practitioner. A chemical composition can be administered at just one concentration, 
particularly where other studies or past work or field experience with the compound have 
indicated that a particular concentration is the one which is most commonly found in the 
30 body. More commonly, the chemical composition will be added in different concentrations 

to cultures of embryoid bodies run in parallel, so that the effects of the concentration 



18 



differences on gene or protein expression and, hence, the differences in toxicity of the 
composition at different concentrations, can be assessed. Typically, for example, the 
chemical composition will be added at a normal or medium concentration, and bracketed by 
twofold or fivefold increases and decreases in concentration, depending on the degree of 
precision desired. 

Where the composition is one of unknown cytotoxicity, a preliminary study is 
conveniently first performed to determine the concentration ranges at which the composition 
will be tested. A variety of procedures for determining concentration dosages are known in 
the art. One common procedure, for example, is to determine the dosage at which the agent 
is directly cytotoxic. The practitioner then reduces the dose by one half and performs a 
dosing study, typically by administering the agent of interest at fivefold or twofold dilutions 
of concentration to parallel cultures of cells of the type of interest. For environmental 
contaminants, the composition will usually also be tested at the concentration at which it is 
found in the environment. For agricultural chemicals, such as pesticides which leave 
residues on foodstuffs, the agent will usually be tested at the concentration at which the 
residue is found, although it will likely be tested at other concentrations as well. 

E. DETECTING ALTERATIONS IN LEVELS OF GENE OR PROTEIN 
EXPRESSION 

1. Detecting Protein Expression Alterations 

Protein expression can be detected by a number of methods known in the art. For 
example, the proteins in a sample can be separated by sodium dodecyl sulphate- 
polyacrylamide gel electrophoresis ("SDS-PAGE") and visualized with a stain such as 
Coomassie blue or a silver stain. Radioactive labels can be detected by placing a sheet of X- 
ray film over the gel. Proteins can also be separated on the basis of their isoelectric point via 
isoelectric focusing, and visualized by staining. Further, SDS-PAGE can be performed in 
combination with isoelectric focusing (usually performed in perpendicular directions) to 
provide two-dimensional separation of the proteins in a sample. Proteins can further be 
separated by such techniques as high pressure liquid chromatography, FPLC, thin layer 
chromatography, affinity chromatography, gel-filtration chromatography, ion exchange 
chromatography, surface enhanced laser desorption/ionization ("SELDI"), matrix-assisted 



19 



laser desorption/ionization ("MALDI"), and, if the sedimentation rates are sufficiently 
different, density gradient centrifugation. Detecting alterations in levels of protein 
expression using these techniques can be accomplished, for example, by running in parallel 
samples from embryoid bodies contacted with a chemical composition whose effect is of 
interest ("test samples") and samples from embryoid bodies cultured under identical 
conditions except for the presence of the chemical composition of interest ("control 
samples"), and noting any differences in the proteins detected and the amount of the proteins 
detected. 

Immunodetection provides a group of useful techniques for detecting alterations in 
protein expression. In these techniques, antibodies are typically raised against the protein by 
injecting the protein into mice or rabbits following standard protocols, such as those taught in 
Harlow and Lane, Antibodies, A Laboratory Manual (Cold Spring Harbor Laboratory, Cold 
Spring Harbor, NY, 1988). The antibodies so raised can then be used to detect the presence 
of and quantitate the protein in a variety of immunological assays known in the art, such as 
ELIS As, fluorescent immunoassays, Western and dot blots, immunoprecipitations, and focal 
immunoassays. Alterations in protein expression can be determined by running parallel tests 
on test and control samples and noting any differences in results between the samples. 
Results of ELISAs, for example, can be directly related to the amount of protein present. 

Tagging provides another way to detect and determine changes in protein expression. 
For example, the gene encoding the protein can be engineered to produce a hybrid protein 
containing a detectable tag, so that the protein can be specifically detected by detection of the 
tag. Systems are available which permit the direct imaging and quantitation of radioactive 
labels in, for example, gels on which the proteins have been separated. Differences in 
expression can be determined by observing differences in the amount of the tag present in 
test and control samples. 

Proteins can also be analyzed by standard protein chemistry techniques. For example, 
proteins can be analyzed by performing proteolytic digests with trypsin, Staphylococcus B 
protease, chymotrypsin, or other proteolytic enzymes. Differences in expression can be 
determined by comparing relative amounts of the digested products. 

One particularly preferred method for determining differences in protein expression is 
mass spectroscopy, or "MS," which provides the broadest profile of the broadest number of 



20 



proteins for the least effort. Moreover, MS permits not only accurate detection of proteins 
present in a sample, but also quantitation. The procedure can be used either by itself, or in 
combination with one or more of the preceding methods based on selective physical 
properties to partition the proteins present in a sample. Partitioning reduces the number of 
proteins of different physical properties in the sample and results in a better MS analysis by 
permitting a comparison of proteins of similar size, electrostatic charge, affinity for metal 
ions, or the like. Thus, for example, the proteins in a sample can be subjected to SDS-PAGE 
and isoelectric focusing, and a resulting spot of interest on the gel can then be subjected to 
MS. In Example 1, below, initial partitioning was performed using a sizing column and a 
second partitioning was performed using SELDL It should be noted that, in the protocol 
followed in Example 1, the proteins with molecular weights smaller than 30 kD were 
analyzed. Alternatively, of course, the higher weight proteins could be analyzed in the 
methods of the invention, and the proteins do not need to be fractionated if the practitioner is 
prepared to analyze all the proteins in a sample or, for example, if a preliminary analysis 
shows that the total number of different proteins in a sample is small enough to be analyzed 
without partitioning. 

Computers attached to the mass spectrometer can also be used to analyze the samples 
to facilitate determination of whether a change in protein expression may be indicative of a 
particular toxicity. For example, the readout from the MS can be used in a "subtractive 
calculation" in which the protein expression in control embryoid bodies is quantitated and 
then subtracted from the quantitated protein expression of embryoid bodies contacted with a 
chemical composition, with only the proteins expressed in greater or lesser quantities than 
those expressed by the control embryoid bodies being shown. This method immediately 
focuses attention on differences in protein expression between a control and a test population. 
Examples of such comparisons are shown in Figures IB and 1C and discussed in detail in 
Example 1, below. 

2. Detecting Gene Expression Alterations 

A number of methods are known in the art for detecting and comparing levels of gene 
expression. 

One standard method for such comparisons is the Northern blot. In this technique, 
RNA is extracted from the sample and loaded onto any of a variety of gels suitable for RNA 



21 



analysis, which are then ran to separate the RNA by size, according to standard methods (see, 
e.g., Sambrook, J., et al, Molecular Cloning, A Laboratory Manual. Cold Spring Harbor 
Laboratory Press, Cold Spring Harbor, NY (2nd ed. 1989)). The gels are then blotted (as 
described in Sambrook, supra), and hybridized to probes for RNAs of interest. The probes 

5 can be radioactive or non-radioactive, depending on the practitioner's preference for 

detection systems. For example, hybridization with the probe can be observed and analyzed 
by chemiluminescent detection of the bound probes using the "Genius System," (Boehringer 
Mannheim Corporation, Indianapolis, IN), following the manufacturer's directions. Equal 
loading of the RNA in the lanes can be judged, for example, by ethidium bromide staining of 

1 0 the ribosomal RNA bands. Alternatively, the probes can be radiolabeled and detected 

autoradiographically using photographic film. 

The RNA can also be amplified by any of a variety of methods and then detected. 

0 For example, Marshall, U.S. Patent No. 5,686,272, discloses the amplification of RNA 

fl sequences using ligase chain reaction, or "LCR." LCR has been extensively described by 
if Landegren et aL 9 Science, 241:1077-1080 (1988); Wu et cd. 9 Genomics, 4:560-569 (1989); 
fll Barany, in PCR Methods and Applications, 1 :5-16 (1991); and Barany, Proc. Natl. Acad. Sci. 

1 USA, 88:189-193 (1991). Or, the RNA can be reverse transcribed into DNA and then 
5 amplified by LCR, polymerase chain reaction ("PCR"), or other methods. An exemplar 

III protocol for conducting reverse transcription of RNA is taught in U.S. Patent No. 5,705,365. 
2(fc Selection of appropriate primers and PCR protocols are taught, for example, in Innis, M., et 

^ al f eds., PCR Protocols 1990 (Academic Press, San Diego CA) (hereafter "Innis et air). 
Differential expression of messenger RNA can also be compared by reverse transcribing 
mRNA into cDNA, which is then cleaved by restriction enzymes and electrophoretically 
separated to permit comparison of the cDNA fragments, as taught in Belyavsky, U.S. Patent 
25 No. 5,814,445. 

Typically, primers are labeled at the 5 T terminus with biotin or with any of a number 
of fluorescent dyes. Probes are usually labeled with an enzyme, such as horseradish 
peroxidase (HRP) and alkaline phosphatase, see, Levenson and Chang, Nonisotopically 
Labeled Probes and Primers in Innis, et al y supra, but can also be labeled with, for example, 
30 biotin-psoralen. Detailed example protocols for labeling primers and for synthesizing 

enzyme-labeled probes are taught by Levenson and Chang, supra. Or, the probes can also be 



22 



labeled with radioactive isotopes. An exemplar protocol for synthesizing radioactively 
labeled DNA and RNA probes is set forth in Sambrook et al, supra. Usually, 32 P is used for 
labeling DNA and RNA probes. A number of methods for detection of PCR products are 
known. See, e.g., Innis, supra, which sets forth a detailed protocol for detecting PCR 
products using non-isotopically labeled probes. Generally, there is a step permitting 
hybridization of the probe and the PCR product, following which there are one or more 
development steps to permit detection. 

For example, if a biotinylated psoralen probe is used, the hybridized probe is 
incubated with streptavidin HRP conjugate and then incubated then incubated with a 
chromogen, such as tetramethylbenzidine (TMB). Alternatively, if the practitioner has 
chosen to employ a radioactively labeled probe, PCR products to which the probe has 
hybridized can be detected by autoradiography. As another example, biotinylated dUTP 
(Bethesda Research Laboratories, MD) can be used during amplification. The labeled PCR 
products can then be run on an agarose gel, Southern transferred to a nylon filter, and 
detected by, for example, a streptavidin/alkaline phosphatase detection system. A protocol 
for detecting incorporated biotinylated dUTP is set forth, e.g., in Lo et al, Incorporation of 
Biotinylated dUTP, in Innis et al, supra. Finally, the PCR products can be run on agarose 
gels and nucleic acids detected by a dye, such as ethidium bromide, which specifically 
recognizes nucleic acids. 

Sutcliffe, U.S. Patent 5,807,680, teaches a method for the simultaneous identification 
of differentially expressed mRNAs and measurement of relative concentrations. The 
technique, which comprises the formation of cDNA using anchor primers followed by PCR, 
allows the visualization of nearly every mRNA expressed by a tissue as a distinct band on a 
gel whose intensity corresponds roughly to the concentration of the mRNA. 

Another group of techniques employs analysis of relative transcript expression levels. 
Four such approaches have recently been developed to permit comprehensive, high 
throughput analysis. First, cDNA can be reverse transcribed from the RNAs in the samples 
(as described in the references above), and subjected to single pass sequencing of the 5' and 
V ends to define expressed sequence tags for the genes expressed in the test and control 
samples. Enumerating the relative representation of the tags from the different samples 



23 



provides an approximation of the relative representation of the gene transcript within the 
samples. 

Second, a variation on ESTs has been developed, known as serial analysis of gene 
expression, or "SAGE," which allows the quantitative and simultaneous analysis of a large 
number of transcripts. The technique employs the isolation of short diagnostic sequence tags 
and sequencing to reveal patterns of gene expression characteristic of a target function, and 
has been used to compare expression levels, for example, of thousands of genes in normal 
and in tumor cells. See, e.g., Velculescu, et aL 9 Science 270:368-369 (1995); Zhang, et al, 
Science 276:1268-1272 (1997). 

Third, approaches have been developed based on differential display. In these 
approaches, fragments defined by specific sequence delimiters can be used as unique 
identifiers of genes, when coupled with information about fragment length within the 
expressed gene. The relative representation of an expressed gene within a cell can then be 
estimated by the relative representation of the fragment associated with that gene. Examples 
of some of the several approaches developed to exploit this idea are the restriction enzyme 
analysis of differentially-expressed sequences ("READS") employed by Gene Logic, Inc., 
and total gene expression analysis ("TOGA") used by Digital Gene Technologies, Inc. 
CLONTECH, Inc. (Palo Alto, CA), for example, sells the Delta™ Differential Display Kit for 
identification of differentially expressed genes by PCR. 

Fourth, in preferred embodiments, the detection is performed by one of a number of 
techniques for hybridization analysis. In these approaches, RNA from the sample of interest 
is usually subjected to reverse transcription to obtain labeled cDNA. The cDNA is then 
hybridized, typically to oligonucleotides or cDNAs of known sequence arrayed on a chip or 
other surface in a known order. The location of the oligonucleotide to which the labeled 
cDNA hybridizes provides sequence information on the cDNA, while the amount of labeled 
hybridized RNA or cDNA provides an estimate of the relative representation of the RNA or 
cDNA of interest. Further, the technique permits simultaneous hybridization with two or 
more different detectable labels. The hybridization results then provide a direct comparison 
of the relative expression of the samples. 

A number of kits are commercially available for hybridization analysis. These kits 
allow identification of specific RNA or cDNAs on high density formats, including filters, 



24 



microscope slides, microchips, and technologies relying on mass spectrometry. For example, 
Affymetrix, Inc. (Santa Clara, CA), markets GeneChip™ Probe arrays containing thousands 
of different oligonucleotide probes with known sequences, lengths, and locations within the 
array for high accuracy sequencing of genes of interest. CLONTECH, Inc.'s (Palo Alto, CA) 
Atlas™ cDNA Expression Array permits monitoring of the expression patterns of 588 
selected genes. Hyseq, Inc.'s (Sunnyvale, CA) Gene Discovery Module permits high 
throughput screening of RNA without previous sequence information at a resolution of 1 
mRNA copy per cell. Incyte Pharmaceuticals, Inc. (Palo Alto, CA) offers microarrays 
containing, for example, ordered oligonucleotides of human cancer and signal transduction 
genes. Techniques used by other companies in the field are discussed in, e.g., Service. R., 
Science 282:396-399 (1998) 
3. Labels 

Both proteins and genes can be labeled to detect the alteration in levels of expression 
in the methods of the invention. The term "label" refers to a composition detectable by 
spectroscopic, photochemical, biochemical, immunochemical, or chemical means. For 
example, useful nucleic acid and protein labels include 32 P, 35 S, fluorescent dyes, electron- 
dense reagents, enzymes (e.g., as commonly used in an ELISA), biotin, dioxigenin, or 
haptens and proteins for which antisera or monoclonal antibodies are available. 

A wide variety of labels and conjugation techniques are known and are reported 
extensively in both the scientific and patent literature, and are generally applicable to the 
present invention for the labeling of nucleic acids, amplified nucleic acids, and proteins. 
Suitable labels include radionucleotides, enzymes, substrates, cofactors, inhibitors, 
fluorescent moieties, chemiluminescent moieties, magnetic particles, and the like. Labeling 
agents optionally include e.g., monoclonal antibodies, polyclonal antibodies, proteins, or 
other polymers such as affinity matrices, carbohydrates or lipids. Detection of labeled 
nucleic acids or proteins may proceed by any of a number of methods, including 
immunoblotting, tracking of radioactive or bioluminescent markers, Southern blotting, 
Northern blotting, or other methods which track a molecule based upon size, charge or 
affinity. The particular label or detectable moiety used and the particular assay are not 
critical aspects of the invention. 



25 



The detectable moiety can be any material having a detectable physical or chemical 
property. Such detectable labels have been well developed in the field of gels, columns, and 
solid substrates, and in general, labels useful in such methods can be applied to the present 
invention. Thus, a label is any composition detectable by spectroscopic, photochemical, 

5 biochemical, immunochemical, electrical, optical or chemical means. Useful labels in the 

present invention include fluorescent dyes (e.g., fluorescein isothiocyanate, Texas red, 
rhodamine, and the like), radiolabels (e.g., 3 H, 125 1, 35 S, 14 C, or 32 P), enzymes (e.g., LacZ, 
CAT, horse radish peroxidase, alkaline phosphatase and others, commonly used as detectable 
enzymes, either as marker gene products or in an ELISA), nucleic acid intercalators (e.g., 

10 ethidium bromide) and colorimetric labels such as colloidal gold or colored glass or plastic 

(e.g. polystyrene, poly-propylene, latex, etc.) beads, as well as electronic transponders (e.g., 
0 U.S. Patent 5,736,332). 

§y It will be recognized that fluorescent labels are not to be limited to single species 

'% organic molecules, but include inorganic molecules, multi-molecular mixtures of organic 
1 § } and/or inorganic molecules, crystals, heteropolymers, and the like. Thus, for example, CdSe- 
feL CdS core-shell nanocrystals enclosed in a silica shell can be easily derivatized for coupling to 
j h a biological molecule. Bruchez et al (1998) Science 281: 2013-2016. Similarly, highly 
tfl fluorescent quantum dots (zinc sulfide-capped cadmium selenide) have been covalently 
| ; y coupled to biomolecules for use in ultrasensitive biological detection. Warren and Nie 
2(P (1998) Science 281: 2016-2018. 

The label is coupled directly or indirectly to the desired nucleic acid or protein 
according to methods well known in the art. As indicated above, a wide variety of labels 
may be used, with the choice of label depending on the sensitivity required, ease of 
conjugation of the compound, stability requirements, available instrumentation, and disposal 
25 provisions. Non-radioactive labels are often attached by indirect means. Generally a ligand 

molecule (e.g., biotin) is covalently bound to a polymer. The ligand then binds to an anti- 
ligand (e.g., streptavidin) molecule which is either inherently detectable or covalently bound 
to a signal system, such as a detectable enzyme, a fluorescent compound, or a 
chemiluminescent compound. A number of ligands and anti-ligands can be used. Where a 
30 ligand has a natural anti-ligand, for example, biotin, thyroxine, and Cortisol, it can be used in 

26 



conjunction with labeled anti-ligands. Alternatively, any haptenic or antigenic compound 
can be used in combination with an antibody. 

Labels can also be conjugated directly to signal generating compounds, e.g., by 
conjugation with an enzyme or fluorophore. Enzymes of interest as labels will primarily be 
hydrolases, particularly phosphatases, esterases and glycosidases, or oxidoreductases, 
particularly peroxidases. Fluorescent compounds include fluorescein and its derivatives, 
rhodamine and its derivatives, dansyl, umbelliferone, fluorescent green protein, and the like. 
Chemiluminescent compounds include luciferin, and 2,3-dihydrophthalazinediones, e.g., 
luminol. 

Means of detecting labels are well known to those of skill in the art. Thus, for 
example, where the label is a radioactive label, means for detection include a scintillation 
counter, proximity counter (microtiter plates with scintillation fluid built in), or photographic 
film as in autoradiography. Where the label is a fluorescent label, it may be detected by 
exciting the fluorochrome with the appropriate wavelength of light and detecting the 
resulting fluorescence, e.g., by microscopy, visual inspection, via photographic film, by the 
use of electronic detectors such as charge coupled devices (CCDS) or photomultipliers and 
the like. Similarly, enzymatic labels may be detected by providing appropriate substrates for 
the enzyme and detecting the resulting reaction product. Finally simple colorimetric labels 
are often detected simply by observing the color associated with the label. Thus, in various 
dipstick assays, conjugated gold often appears pink, while various conjugated beads appear 
the color of the bead. 

F. CORRELATING MOLECULAR PROFILES WITH TOXICITIES 

The invention contemplates multiple iterations of compiling a library of molecular 
profiles by contacting test embryoid bodies with an ever-widening group of chemical 
compositions having predetermined toxicities. The toxicities and biological effects of many 
chemical compositions are already known through previous animal or clinical testing. Any 
such information is carefully noted along with the alterations of gene or protein expression in 
embryoid bodies. As the data from tests on a number of chemical compositions, or agents, is 
gathered, it is assembled to form a library. Separate libraries can be maintained for each type 
of toxicity; preferably, a single database can be maintained recording the results of all the 



27 



tests conducted and any available toxicity information on the agents to which the embryoid 
bodies were exposed. Preferably, biological effects are also noted. Past experience has 
indicated that biological effects often become associated with, or markers for, particular 
toxicities as the biology of the toxicity becomes better understood. 

The invention contemplates that each iteration of contacting test embryoid bodies 
with a chemical composition will generate a pattern of gene or protein expression, or both, 
characteristic for that chemical composition. The determination of the alteration in gene or 
protein expression of a reasonably large number of chemical compounds of similar toxicity is 
desirable so that patterns of gene or protein expression, or both, associated with that toxicity 
can be determined. Changes in gene or protein expression patterns in EB cells that are 
common to classes of drugs that have similar toxicities will serve as surrogate molecular 
profiles useful for recognizing compounds that are likely to have related biology and 
toxicities. It is the correlation of these alterations in gene or protein expression and toxicities 
that gives the invention its predictive power with respect to previously untested compounds. 

The correlation of patterns of gene or protein expression with toxicities can be 
performed by any convenient means. For example, visual comparisons of patterns can be 
performed to determine patterns associated with different types of toxicities. More 
conveniently, the correlation can be done by computer, using one of the database programs 
discussed in the previous section. Preferably, the correlation is performed by a computer 
using a neural network program, since neural network programs are specifically designed for 
pattern recognition. Once a correlation of expression markers which are biomarkers for a 
particular toxicity has been made, a comparison can be made, again conveniently by 
computer, of known patterns to the pattern of gene or protein expression induced by a new or 
unknown chemical composition to provide the closest matches of expression. The patterns 
can then be reviewed to predict the likely toxicity of the new or unknown chemical. 

G. TYPING AND RANKING TOXICITIES OF TEST CHEMICAL 
COMPOSITIONS 

A molecular profile of a test chemical composition can be established by detecting 
the alterations in gene or protein expression in embryoid bodies contacted by the test 
chemical composition as described in previous sections. Once the molecular profile of the 



28 



test composition is determined, it can be compared to that of a chemical composition with 
predetermined toxicities or, preferably, to a library of molecular profiles of chemical 
compositions with predetermined toxicities. The outcome of such comparison provide 
information for one to predict the likelihood of whether the test composition is toxic, what 
5 type of toxicities, and how toxic it would be as compared to the other known toxic 

compositions. 

For the purpose of practicing the invention, the predictions of toxicity of the test 
composition based on its molecular profiles in EB cells does not have to be 100% accurate. 
To have a major positive impact on the efficiency and costs of drug development, one only 
10 has to modestly increase the probability that the less toxic and thus more successful drug 

candidates are, for example, on the top half of a prioritized list of new drug leads. 

As noted in previous sections, alterations in gene or protein expression in embryoid 
;J3 bodies exposed to a chemical composition can be detected by any of a number of means 
Jf known in the art. Protein expression determined by MS is particularly convenient for such 
1 5p comparisons since the output data is typically fed directly into a computer connected to the 
ft j mass spectrometer and is immediately available for a variety of calculations. If the 
m alterations are susceptible to graphical representation, as when MS is used as the means of 
0 detection, a direct comparison can be made of the effect of the chemical composition on the 
1 1 expression of proteins compared to the control embryoid bodies. If the alterations are 
2(j^ detected by, for example, an ELISA, which produces a numerical readout, then the numerical 
N readouts can be used to quantitate the expression of the protein. For gene expression, 
Northern blots can be correlated to the amount of RNA present for each RNA probed. 
Where gene expression is detected by hybridization arrays, the pattern of hybridization for 
nucleic acids from the test and control embryonic bodies provides a basis for comparison. 
25 The comparison of molecular profiles can be done by a number of means known in 

the art. Usually, the graphs resulting from the calculations can be stored, for example, in file 
folders or the like, and examined visually to discern common patterns of expression 
compared to the control, as well as differences. Conveniently, however, the data can be 
stored on and compared by a computer. Programs are available, for example, to compare 
30 mass spectrometry data. Figures IB and 1C, for example, demonstrate the use of 

"subtractive calculation" and graphical representation to compare protein expression in the 



29 



control embryoid bodies ("control samples") against that of the embryoid bodies contacted 
with either of two chemical compositions ("test samples"). In this comparison, the amount of 
each protein expressed by the control samples is subtracted from the amount expressed by the 
test samples. The control sample value is represented by a horizontal line, and any protein 
expressed in a different amount is represented as a line above or below the line (representing 
positive and negative amounts compared to the control, respectively), with the height of the 
line designating the amount by which the expression of the test sample is different from that 
of the control. This method focuses attention on the differences in protein expression. In a 
like manner, the program can also be used to compare the expression of two or more test 
samples so that any differences in expression patterns can be readily discerned. It is expected 
that the more similar the pattern of expression, the more similar will be the effect, and the 
type of toxicity, of the two agents. 

Another form of comparison is shown in Figures 2, 3, and 4. These figures 
graphically depict the small nuclear, small cytoplasmic, and large cytoplasmic proteins 
expressed by control samples and by test samples exposed to one of two chemical 
compositions, as well the amount of the protein expressed by the samples. These graphs can 
be compared visually, and the proteins and the amounts expressed recorded manually. 
Preferably, the results are placed into a computer database, with information about the known 
toxicities of the chemical compositions recorded in searchable data fields. Entries of data 
from other forms of detecting alterations in protein or gene expression can also be reviewed 
and recorded manually or in a computer database. For example, the values from an ELISA, 
or the proteins identified on a Western blot can be recorded to identify the types and amounts 
of proteins expressed in control and test samples. Similarly, the patterns on a Northern blot, 
or the hybridization pattern on an oligonucleotide array, can be recorded to identify the gene 
expression of control and test samples. The information can be kept manually, but preferably 
is maintained in a computer searchable form. 

Standard database programs, such as Enterprise Data Management (Sybase, Inc., 
Emeryville, CA) or Oracle8 (Oracle Corp., Redwood Shores, CA) can be used to store and 
compare information. Alternatively, the data can be recorded, or analyzed, or both, in 
specifically designed programs available, for example, from Partek Inc. (St. Charles, MO). 



30 



Additionally, companies selling integrated analytical systems, such as mass 
spectrometers, provide with the machines integrated software for recording results. Such 
companies include Finnigan Corp. (San Jose, CA), Perkin-Elmer Corp. (Norwalk CT), 
Ciphergen Biosystems, Inc. (Palo Alto CA), and Hewlett Packard Corp. (Palo Alto, CA). 
Similarly, companies such as Incyte Pharmaceuticals, Inc. (Palo Alto CA) providing 
oligonucleotide hybridization services maintain proprietary image recognition algorithms to 
record and analyze the scanned images of hybridization arrays. 

In a preferred embodiment, the data can be recorded and analyzed by neural network 
technology. Neural networks are complex non-linear modeling equations which are 
specifically designed for pattern recognition in data sets. One such program is the 
NeuroShell Classifier™ classification algorithm from Ward Systems Group, Inc. (Frederick, 
MD). Other neural network programs are available from, e.g., Partek, Inc., BioComp 
Systems, Inc. (Redmond WA) and Z Solutions, LLC (Atlanta, GA). 

H. ADAPTING ARRAY READERS 

In one embodiment, the invention relates to the formation of arrays of hybridized 
oligonucleotides or of bound proteins to detect changes in gene or protein expression, 
respectively. Such arrays can be scanned or read by array readers. 

Typically, the array reader will have an optical scanner adapted to read the pattern of 
labels on an array, such as of bound proteins or hybridized oligonucleotides, operably linked 
to a computer which has stored on it, or accessible to it (for example, on an external drive or 
through the internet) one or more data files having a plurality of gene expression or protein 
expression profiles of mammalian embryoid bodies contacted with known or unknown toxic 
chemical compositions. The array reader can, however, be adapted with a detection device 
suitable to "read" labels that can not be read optically, such as electronic transponders. 

I. USE IN HIGH THROUGHPUT SCREENING 

The methods of the invention can be readily adapted to high throughput screening. 
High throughput ("HTP") screening is highly desirable because of the large number of 
uncharacterized compounds already developed in the larger pharmaceutical companies, as 
well as the flood of new compounds now being synthesized by combinatorial chemistry. 



31 



Using the invention, hundreds of chemical compositions can be tested on embryoid bodies 
and the resulting alterations in gene or protein expression, or both, compared to toxicities of 
known chemical compositions to predict the type and possibly the degree of toxicity the new 
compounds possess. Those compositions with acceptable toxicity profiles can then be 
considered for further levels of testing. 

HTP screening can be facilitated by using automated and integrated culture systems, 
sample preparation (protein or RNA/cDNA), and analysis. These steps can be performed in 
regular labware using standard robotic arms, or in more recently developed microchip and 
microfluidic devices, such as those developed by Caliper Technologies Corp. (Palo Alto, 
CA), described in U.S. Patent 5,800,690, by Orchid Biocomputer, Inc. (Princeton, NJ), 
described in the October 25, 1997 New Scientist, and by other companies, which provide 
methods of automated analysis using very low volumes of reagents. See, e.g., McCormick, 
R., et aL, Anal Chem. 69:2626-2630 (1997); Turgeon, M., Med Lab. Management Rept, 
Dec. 1997, page L 

EXAMPLES 

Example 1. Selecting chemical compounds for toxicity screening 

Compositions that fall into particular categories of toxicity are used to establish 
molecular profiles and compile libraries for particular toxicities. Table 1 lists a number of 
compositions that are known to be toxic to certain tissues or organs or during developmental 
stages. In particular, those compositions causing liver toxicities are assessed for their 
molecular profiles by determining alterations of gene or protein expression patterns in 
embryoid bodies contacted by each composition. A library comprising molecular profiles of 
compositions having liver toxicities is therefore compiled. Those compositions causing 
cardiorvascular toxicities are similarly assessed for their molecular profiles and a library 
compiled. In addition, molecular profiles and library thereof for compositions having 
toxicities on central nervous system and for compositions having developmental toxicities 
are similarly established using the embryoid body system. The experimental procedures as 
described above in general, and in more detail in the following examples, are followed to 
compile the molecular profiles and libraries for compositions with particular type of 
toxicities. 



32 



Drags with known or suspected of having activities against particular diseases can be 
used to establish molecular profiles and libraries for toxicity assessment. Antineoplastics 
drugs with similar toxicities, for example those listed in Table 1, can be used to compile 
molecular profiles by determining the alterations in gene or protein expression patterns in 
embryoid bodies exposed to these drugs. Similarly, antibiotics with similar toxicities can 
also be assessed for their alterations in gene or protein expression patterns in embryoid 
bodies. Also used are drugs controlling diabetes, drags for lowering lipid levels, or anti- 
inflammatory drags. Once a composite library comprising molecular profiles of specific type 
of drugs having similar toxicities is established, it can be used to screen for new drug leads of 
the similar type for their potential toxicities. Again, the experimental procedures as 
described above in general, and in more detail in the following examples, are followed for 
compiling molecular profiles and libraries, and for typing/ranking toxicities of new drug 
leads. 

Example 2. Establishing protein profiles for chemical agents relating to liver toxicities 

This Example demonstrates the culturing of embryoid bodies, the exposure of the 
embryoid bodies to different chemical agents having liver toxicities, and the determination of 
changes in protein expression in the embryoid bodies. 

Five thousand CCE embryonic stem cells (Robertson, E., et aL, Nature 323:445-448 
(1986), were maintained and harvested according to Keller (Keller, G., et al, Mol. Cell Biol, 
13:473-486 (1993). Briefly, the cells were cultured in 5 mis of IMDM medium, 20% FCS, 
ascorbic acid (50 jig/ml), and monothioglycerol (2.6 x 10" 5 v/v) at 37°C with 6% C0 2 . On 
day 2, troglitazone, a drug marketed for the control of diabetes which has shown rare but 
severe liver toxicity, was added at a final concentration of 20 fiM to one group of plates 
(group "A") containing embryoid bodies. On that same day, erythromycin estolate (Sigma 
catalog E8630), a form of erythromycin with known liver toxicity, was added to a second 
group of plates (group "B") at a final concentration of 50 jaM. A third group of plates 
containing embryoid bodies (group "CI") was cultured without any added drugs to serve as a 
control. Additionally, plates containing only tissue culture medium (group "C2") were 
cultured alongside of those containing embryoid bodies as a control for degradation of 
proteins in the culture medium. After six days, and again at nine days, the cultures were 

33 



harvested, the cells washed twice with PBS, and lysed in PBS, 0.5% Triton X100 for 10 
minutes on ice. The nuclei were pelleted, and the supernatant removed and stored at -80°C 
until analysis. The nuclei were lysed in PBS with 0.2% SDS and dounce homogenized to 
shear the DNA. The insoluble material was pelleted and the nuclear lysates stored at -80°C 
until analyzed. Cytoplasmic and nuclear lysates were also taken on day zero prior to 
exposure to any test chemical compositions to serve as additional controls. 

The lysates and medium samples were diluted 3 fold in buffer containing 50 mM 
Tris-HCl at pH 8, and 0.4 M NaCl. Aliquoted samples of diluted lysate or medium were 
placed in a sizing spin column that fractionated the sample with a 30 kD cutoff and 
equilibrated in 50 mM Tris-HCl, pH 8 and 50 mM NaCl. The column was spun at 700 g for 
3 minutes for each fraction. Four fractions of 25 jliL were collected for each column using 
the column equilibrated buffer. 

The samples were partitioned by surface enhanced laser desorption/ionization 
("SELDI"), and proteins were detected by mass spectroscopy. SELDI permits proteins to be 
captured on a surface of choice, which can then be washed at selected stringency, to permit 
fractionation according to desired characteristics such as affinity for metal ions of the surface 
used for capture. 

Ciphergen normal phase chips (Ciphergen Biosystems, Palo Alto, CA) were used to 
partition the proteins in the fractions generated by the spin columns. One jjL aliquots of each 
fraction were deposited on a spot on the chip, and the sample was air dryed at room 
temperature for 5 minutes. A mixture of 0.5 \xL of saturated sinapinic acid ("SPA") in 50% 
acetonitrile with 0.5% trifluroacetic acid ("TFA") was applied to each spot. The chip was 
again permitted to air dry for 5 minutes at room temperature, and a second aliquot of the SPA 
mixture was applied. 

Chips were read by the Ciphergen Protein Biology System 1 reader. Auto mode was 
used for data collection, at the SELDI quantitation setting. Two sets of protein profiles were 
collected, one at low laser intensity (at 15 with filter out) and one at high laser intensity (at 
50 with filter out), detector set at 10. An average of 15 shots per location on the same sample 
spot were made. Protein profiles from different lysates were compared using SELDI 
software (Ciphergen Biosystems, Palo Alto, CA). This program assumes two proteins with a 
molecular weight within 1% of each other are the same. It then quantitates the results, 



34 



compares the test samples against the control samples, and prints a graph showing the 
amount of each protein in the control as a horizontal line, with any reduction or the excess in 
the amount of each protein in the test sample compared to the amount of that protein in the 
control sample as a line below or above the line representing the control 

The results of these analyses for the day 6 embryoid bodies are shown as Figures 1 
through 4. One portion of the results of this analysis, the differences in nuclear proteins 
expressed by the embryoid bodies, is shown in Figure 1. The top panel, panel 1 A, is a half- 
tone reproduction of the readout from the mass spectrometer. Viewing the sheet from along 
the long axis, the top band, is the mass spectrum for the control, the embryoid bodies grown 
in the absence of either of the test chemical compositions, the middle band is the spectrum 
for the embryoid bodies grown in the presence of added troglitazone, and the bottom band of 
Figure 1 A shows the mass spectrum of nuclear proteins expressed by embryoid bodies 
exposed to erythromycin estolate. 

Figures IB and 1C graphically depict differences in protein expression level between 
embryoid bodies contacted with one of the test chemical compositions ("test embryoid 
bodies") and control embryoid bodies grown in standard tissue growth medium without 
added chemical compositions. These panels present computational subtractions of identical 
proteins between the respective test embryoid bodies and the control embryoid bodies to 
indicate only those proteins which are significantly different in expression between the test 
and the control embryoid bodies. Each bar represents a single protein and the length of the 
bar represents the amount of protein expressed by the embryoid bodies exposed to the test 
composition compared to the amount expressed by the control embryoid bodies. A bar above 
the center line indicates that the test embryoid body expressed more of that protein than did 
the control embryoid bodies; a bar below the line indicates that the test embryoid body 
expressed less of that protein. 

Figure IB shows the differences in the nuclear proteins expressed by embryoid bodies 
grown in the presence of troglitazone compared to control embryoid bodies. Figure 1C 
shows the differences in the nuclear proteins expressed by the embryoid bodies grown in the 
presence of erythromycin estolate and the control. (Both the test and the control embryoid 
bodies were at day 6 of development.) Reading Figures IB and 1C from the left, the first bar 
encountered is above the line at the same position for both Figures, but the height of the bar 



35 



is much greater in Figure 1C. This indicates that both groups of test embryoid bodies 
expressed more of this protein than did the control, but that the bodies contacted with 
erythromycin estolate expressed considerably more than did bodies contacted with 
troglitazone. 

Continuing along the X, or molecular weight, axis of Figure 1C, the next four bars 
encountered also have a counterpart in Figure IB. Moreover, in each of the Figures, the bars 
representing the same three proteins are below the line, whereas the bar for the same fourth 
protein is above the line. Once again, the height of the lines differs between Figures 1C and 
IB. Thus, for the first 5 nuclear proteins detected, the embryoid bodies contacted with 
troglitazone and with erythromycin estolate displayed the same pattern of protein expression, 
but at different levels of expression. Each of these proteins, and the overall expression 
pattern, would be a candidate for inclusion in a profile indicating that an unknown chemical 
composition, such as a new potential therapeutic, had some liver toxicity. Conversely, the 
first protein detected in Figure 1C to the right of the 4000 Daltons molecular weight line does 
not have a counterpart (or at least a counterpart in terms of being expressed at a level 
different from that of the control bodies) in Figure IB. This protein would therefore not be 
considered a protein that demonstrated a common pathway of liver toxicity of both 
troglitazone and erythromycin estolate. Depending on its correlation with expression 
pathways of other hepatic toxins, it might, however, be associated with liver toxicity. Similar 
analyses can be made for the other proteins depicted on the two graphs. 

A further way to present an analysis of the differences in protein expression can be 
seen in Figure 2. Figure 2 compares also the expression of small nuclear proteins in the three 
embryoid body groups described above. In these graphs, each bar in a panel represents a 
single protein, but the length of the bar represents the relative amount of protein expressed, 
rather than a comparison of the amount expressed compared to the control embryoid bodies. 
In Figure 2, the top panel, 2 A, graphs the level of protein expression, as determined by mass 
spectroscopy, in the embryoid bodies not exposed to chemical compositions in addition to 
those in a standard tissue culture medium. The middle panel, 2B 5 shows the level of 
expression of proteins of embryoid bodies exposed to troglitazone. And the bottom panel, 
2C, shows the level of expression of embryoid bodies contacted with erythromycin estolate. 
In these panels, the expression level of the protein, plotted on the Y axis as a relative value, is 



36 



plotted against the molecular weight, plotted on the X axis. A visual comparison of the 
panels reveals that some of the proteins expressed by the embryoid bodies exposed to the two 
drags tested are the same, although perhaps at different levels of expression, and that others 
are different, and that both show a different pattern of expression than do the control 
embryoid bodies not exposed to either drug. 

Figure 3 shows the level of expression of small cytoplasmic proteins in the same three 
groups of embryoid bodies as those discussed in the preceding paragraph. The panels are 
arranged in the same order as in Figure 2. Once again, the expression level of the protein for 
each group, plotted on the Y axis is plotted against the molecular weight of the proteins, 
plotted on the X axis. Once again, a visual comparison of the panels reveals that some of the 
proteins expressed by the embryoid bodies exposed to the two drugs tested are the same, 
although perhaps at different levels of expression, and that others are different. 

Similarly, Figure 4 sets forth a graphical analysis of the large cytoplasmic proteins 
expressed by the same groups of embryoid bodies discussed above. Once again, the level of 
expression determined by the mass spectrometry is plotted on the Y axis, while the molecular 
weight is plotted on the X axis. Once again, clear similarities, and clear differences, can be 
observed between the protein expression patterns of the embryoid bodies exposed to the test 
chemical compositions, and between those protein expression patterns and that of the 
embryoid bodies grown without exposure to either of the test chemical compositions. 

It is clear from these figures that the two drugs induce complex and unique protein 
expression patterns. Some proteins are expressed in smaller amounts (or "down regulated") 
compared to the protein expression in the control embryoid bodies, and others are expressed 
in higher amounts (or "up regulated") compared to the controls. Additionally, these two 
chemical compositions affect some of the same proteins and thus share common sub- 
patterns. 

For example, in Figure 2C, to the right of the line denoting a molecular weight of 
2500 Daltons, there is a tall line, over 15 units on the Y axis, designating a strongly 
expressed protein. Following the line up to panels 2B and 2A, one can see that that same 
protein is expressed at high levels in both the embryoid bodies contacted with troglitazone 
and in the control embryoid bodies not contacted with either drug. This protein, therefore, is 
highly expressed in embryoid bodies at the point in development at which the samples were 



37 



taken, although there is some variation in level of expression. Continuing to the right in 
panel 2C and making the same comparisons, however, the next protein present is also 
present, in approximately the same amount, in the embryoid bodies exposed to troglitazone, 
but is not expressed at all by the control embryoid bodies. Thus, this protein is a candidate 
for differentiating chemical compositions with liver toxicity from other compositions and 
other kinds of toxicity. 

Example 3. Screening of anti-cancer drugs for tissue and organ toxicities 

This example illustrates using the EMBRYOID BODY system for screening 
anti-cancer agents for their tissue or organ toxicities. 

Compounds and drugs (both anti-cancer and therapeutic) that have known 
toxicities and biology endpoints in humans and/or animals are selected for compiling their 
gene or protein expression profiles in embryoid bodies. In addition, compounds are selected 
with related known mechanisms of activities and with regard to compounds that have been 
used in previous studies to correlate clinical outcomes with human in vitro cell culture 
effects. Table 2. 



38 



TABLE 2 



Toxicities 



Drugs 


Dev 


Liver CV GI 


CNS 


Renal 


Blood 


Mechanism 


chloroquinoxaline 




+ 




+ 




? 


sulfonamide 














didernnin B 












? 


cyclophosphamide 












alkylator 


bizelesin 










+ 


alkylator 


carbopiatin 




+ 






+ 


alkylator 


cisplatin 




+ 




+ 


+ 


alkylator 


oxaliplatin 






+ 






alkylator 


ecteinascidin 743 










+ 


alkylator 


penclomedine 






+ 






alkylator 


methotrexate 


+ 










anti-metabolite 


fuzarabine 










+ 


anti-metabolite 


fludarabine 










+ 


anti-metabolite 


flavopiridol 




+ 








CdK inhibitor 


doxorubicin 




+ 








DNA intercalator 


amonafide 












DNA intercalator 


daunorubicin 




+ 






+ 


DNA syn inhib 


gemcitabine 




+ 






+ 


DNA syn inhib 


etoposide 










+ 


DNA syn inhib 


deoxyspergualin 












immunosuppression 


camptothecin 












topo-I inhibitor 


9 aminocamptothecin 










+ 


topo-I inhibitor 


topotecan 










+ 


topo-I inhibitor 


merbarone 








+ 




topo-II inhibitor 


dolastatin 10 










+ 


tubulin inhibitor 


taxol 










+ 


tubulin inhibitor 


vinblastine 


+ 








+ 


tubulin inhibitor 


vincristine 


+ 










tubulin inhibitor 


vindesine 


+ 








+ 


tubulin inhibitor 


vinorelbine 


+ 








+ 


tubulin inhibitor 


"Dev" = developmental 


"GI = gastro-intestinal 


"CV" 


= cardiovascular 


U CNS" = central nervous system 



a. Establishing gene expression profiles 

The gene expression pattern of a selected compound is measured and quantified using 
cDNA microarrays and is normalized with cellular differentiation. The gene expression 
pattern of the compound is compared with a control EB culture not exposed to the compound 
or, where appropriate, EB cultures treated with related drugs with similar function or dose 
limiting toxicity. By compiling the gene expression profiles for a number of anti-cancer 
agents having similar or related toxicities, common alterations in gene expression are 
discerned and correlated with the toxicities, and are used as surrogate profiles for assessing 
the toxicities of test anti-cancer drug candidates. 



39 



The cDNA microarray can be any one of many kinds that are known and available in 
the art, for example, as described in Shalon et al (1996), Genome Res 6:639-645. cDNA 
microarrays allow for the simultaneous monitoring of the expression of thousands of genes, 
by direct comparison of control and chemically-treated cells. 3' expressed sequence tags 
(ESTs) are arrayed and spotted onto glass microscope slides at a density of hundreds to 
thousands per slide using high speed robotics. Fluorescent cDNA probes are generated from 
control and test RNAs using a reverse transcriptase reaction with labeled dUTP using fluors 
that excite at two different wavelengths, i.e. Cy3 and Cy5, which allows for the hybridization 
of both the control and test RNA to the same chip for direct comparison of relative gene 
expression in each sample. The fluorescent signal is detected using a specially engineered 
scanning confocal microscope. A collection of 15,000 sequence verified human clones and 
8700 mouse clones can be used in making cDNA microarrays. These microarrays are ideal 
for the analysis of gene expression patterns in EB cultures treated with a variety of agents. 

Briefly, RNAs are isolated from control and treated EB cells. Total RNA are 
prepared using the RNAeasy kit from Qiagen. Subsequently, RNA are labeled either with 
Cy3 or Cy5 dUTP in a single round of reverse transcription. The resultant labeled cDNAs 
are mixed in a concentrated volume and hybridized to the arrays. Hybridizations is incubated 
overnight at 65°C in a custom designed chamber that prevents evaporation. Following 
hybridization, the chip is scanned with a custom confocal laser scanner that will provide an 
output of the intensity of each spot in the array for both the Cy3 and Cy5 channels. The data 
is then analyzed with a software package that contains additional extensions. These 
extensions allow for the integration of a signal across each spot, normalization of the data to 
a panel of designated housekeeping genes, and statistical calculations to generate a list of 
genes whose ratios are outliers, or significantly changed by the treatment. In addition to the 
image analysis software, informatics packages such as Spot-Fire and GeneSpring, both are 
commercially available, are used to allow clustering and analysis of genes in multiple 
experiments across dose and/or time. cDNA microarray technology, in general, is still being 
validated as a viable technique for providing quantitative data. While the ratio of red/green 
provides good qualitative data on the relative level of expression of a gene in one population 
versus the other, it is not an absolute value of the level of induction/down regulation of that 
gene. Each pair of samples on the arrays are hybridized in triplicate. Outliers that are 



40 



consistently induced or suppressed in two of the three hybridization experiments are further 
validated by a traditional RNA quantitation method, such as Northern blot or RT-PCR. 

Each drug is tested at least three times on separate EB cultures for its effects on 
growth, differentiation and RNA expression. Cell counts (growth), colony counts 
(differentiation) and RNA levels (cDNA microarrays) are averaged for the three of more 
experiments and the mean and SEM determined. All results are normalized using 
approximately 15 "house keeping" genes. This allows a quantitative comparison of the 
effects of the test drugs to control compounds that are not toxic in humans or animals. 
Statistical comparisons provide information for determining whether a given drug affects EB 
cells gene expression compared to control drugs or non-treated cells and for determining 
whether a change in RNA in the cells is relevant. 

b. Establishing protein expression profiles 

The protein expression profiles of the selected anti-cancer drugs are established using 
Ciphergen' s SELDI mass spectroscopy (MS)-TOF system, as described in Example 2. Total 
cell lysates from harvested EB cultures are prepared in either 0.1% SDS or Triton-XlOO 
(0.5%) and directly applied to protein array chips using manufacture's protocols. Each chip 
can analyze two drugs in triplicate. After working out the stringency conditions and 
experimental replications, on average 6 ProteinChips™ per test compound are used. 

The Ciphergen technology allows for the proteins in the sample to be captured, 
retained and purified directly on the chip. The proteins on the microchip are then analyzed 
by (SELDI). This analysis determines the molecular weight of proteins in the sample. An 
automatic readout of the molecular weights of the purified proteins in the sample can then be 
assessed. Typically this system has a CV of less than 20%. The Ciphergen data analysis 
system normalizes the data to internal reference standards and subtracts the readout of 
proteins found in control cells from those in drug treated cells. This data analysis reveals 
protein expression stimulated by the drugs as well as proteins only found in the control cells 
whose expression is inhibited by the drug. The analysis provides a qualitative readout of 
protein expression between a control and treated group. Analysis of multiple samples 
provides an average fold change in protein expression and a relative measure of variability. 
This can be represented as a mean + SEM which can provide a statistical measure of the 



41 



protein changes. This analysis is used to determine whether drugs that induce similar forms 
of toxicity in humans cause similar changes in protein expression in EB cells. Each drug is 
analyzed on at least 3 separate groups of ES cells. 

All publications and patent applications cited in this specification are herein 
incorporated by reference as if each individual publication or patent application were 
specifically and individually indicated to be incorporated by reference. 

Although the foregoing invention has been described in some detail by way of 
illustration and example for purposes of clarity of understanding, it will be readily apparent 
to those of ordinary skill in the art in light of the teachings of this invention that certain 
changes and modifications may be made thereto without departing from the spirit or scope of 
the appended claims. 



42